图像分割算法在医学图像中的应用综述
Review of the application of image segmentation algorithm in medical images
孙淑婷 1刘铖枨 1周广茵 1韩锐 1陈立超 1羊月祺 2许玥2
作者信息
- 1. 南京医科大学生物医学工程与信息学院,南京 210000
- 2. 南京医科大学第一附属医院(江苏省人民医院)临床医学工程处,南京 210029
- 折叠
摘要
医学图像分割是计算机辅助诊断领域的一项关键技术,其主要任务是将特定的器官、组织或异常区域从图像中准确地识别出来.但是医学图像的质量易受到其复杂纹理和成像设备限制(如噪声和边界不清晰)的影响,故传统的医学图像分割方法已难以满足现实临床需求.随着深度学习技术的进步,基于这一领域的算法已经取得了显著的进展.本文首先回顾了七种传统的医学图像分割策略,并重点介绍了两种当前主流的深度学习方法:全卷积神经网络和U-Net,最后文章探讨了目前深度学习技术所面临的挑战及其可能的解决策略.
Abstract
Medical image segmentation is a key technology in the field of computer-aided diagnosis,whose main task is to accurately identify specific organs,tissues,or abnormal areas from the image.However,the quality of medical images is easily affected by their complex textures and imaging equipment limitations(such as noise and unclear boundaries),so traditional medical image segmentation methods are no longer able to meet practical clinical needs.With the advancement of deep learning technology,algorithms based on this field have made significant progress.This article first reviews seven traditional medical image segmentation strategies and focuses on two current mainstream deep learning methods:fully convolutional neural networks and U-Net,finally,the article also explores the challenges faced by these deep learning technologies and their possible solutions.
关键词
深度学习/医学图像分割/全卷积神经网络/U-NetKey words
Deep learning/Medical image segmentation/Fully convolutional neural network/U-Net引用本文复制引用
基金项目
江苏省高等学校大学生创新创业训练计划项目/南京医科大学大学生创新创业训练计划项目(202310312041Z)
出版年
2024
NETL
NSTL
万方数据